The CRESTA project has just published a new White Paper- The Exascale Development Environment State of the Art and Gap Analysis. You can download the paper and some of our other White papers here.
Here is some information on the White Paper:
The development and implementation of efficient computer codes for exascale supercomputers will
crucially depend on a combined advancement of all development environment components. This
white paper presents the state of the art of programming models, compiler technologies, run-time
systems, debuggers, correctness checkers, and performance monitoring tools and it identifies the
common challenges and problems that need to be solved before the exascale era.
The main focus of this white paper is on emerging and novel technologies in programming models
and tools. Together with the traditional approaches, the white paper presents the PGAS parallel
programming models and new approaches for programming accelerators, such as OpenACC. It is
important that these emerging programming models can be combined with traditional ones for
their uptake on exascale supercomputer. For this reason, we discuss in detail the interoperability of
different programming approaches. Because we recognize that hand-optimization of parallel codes
will be significantly more complex on exascale machines, we present recent progress in software
frameworks for automatic tuning and run-time systems to schedule processes on million of computing
units. Finally, an overview of the state of the art in parallel debuggers, correctness checkers and
performance monitoring and analysis tools is presented focusing on which approaches can provide
scalability on exascale machine.
After discussing the state of the art in the field, we analyze the two main common challenges
for the developments environment on exascale supercomputers. First, all the components of the
programming environments will deal with unprecedented amount of data coming from executing/
debugging/scheduling/monitoring codes running on million computing units, and they will be
required to provide responsiveness and interactivity but still introducing minimal overhead. Second,
programming tools will need to provide support for novel programming models, such as PGAS, and
hardware accelerators, such as GPU and Intel MIC, that will become more and more common on
This white paper provides an overview of different approaches with exascale potential and indicates
the progress that are needed to fill the existing gap between petascale and exascale development
environment technologies. The results of this white paper guide the current work on development
environment in the CRESTA project.
Researchers at University College London (UCL) modelling intracranial blood flow have been able to reach a major milestone by using development tools from Allinea Software. A massive simulation using 50,000 processor cores on the UK’s largest supercomputer, ARCHER, became possible after solving an application crash that only occurred at this high scale. UCL’s HemeLB software applies computational fluid dynamics to model blood flow around cerebral vessels and simulate pressure at points of weakness such as aneurysms. Using data from MRI scans of a patient’s blood vessels, the group anticipates that simulations will one day help to decide the best clinical option for individual patients. The complexity of the simulations requires the UCL team to use ARCHER, the UK’s flagship Cray XC30 system, which is managed by EPCC on behalf of EPSRC and other UK research councils.
The UCL team used Allinea Software’s performance profiling tool, Allinea MAP, to increase the performance of HemeLB on the system. Having already improved performance on some test cases by over 25% they wanted to run a larger simulation.
“We’d never been able to look at this many cores – and get a clear view of how the time was being used – we were keen to see it in Allinea MAP,” UCL Post-doctoral Researcher Derek Groen said. However, the application crashed when using 50,000 processor cores, which stopped them in their tracks. “The crash was totally unexpected. I didn’t know how I would diagnose or fix it at that scale – it was beyond anything I had tried to do before.” says Groen. “Allinea Software helped us straight away – they knew that if we could run the simulation with their debugger, we would find the problem.”
The UCL team and Allinea Software were brought together by the EU CRESTA project – which is preparing applications for future extreme-scale computing. Running applications at extreme scales poses challenges, which Allinea Software’s tools help to solve. Allinea DDT is the only parallel debugger that handles that scale of the problem – which was resolved quickly by debugging all 50,000 application processes simultaneously. “Getting HemeLB to scale to 50,000 ARCHER cores is a real achievement”, says Professor Peter Coveney, Director, Centre for Computational Science, UCL.
“We are thankful for the productive collaborations we enjoy with Allinea Software that have allowed us to reach these intoxicating heights, which are enabling us to study hemodynamics within the Circle of Willis for the first time.” “Working alongside the HemeLB team has been a great experience for us,” says David Lecomber, CEO of Allinea Software. “It helps us to understand the issues they face better. It is inspiring to see the impact HPC could make to lives in the future – and we’ve seen at first-hand the impact on progress that the challenges make. Tools designed to handle scale are changing the way developers think about those hard challenges.”
“Seeing HemeLB used at this scale on ARCHER is very impressive, and highlights the success of the co-design approach used by UCL and Allinea Software on the CRESTA project”, says Dr Lorna Smith, CRESTA project manager and Deputy Director of the ARCHER CSE team. “This achievement with Allinea Software is a great example of the importance of providing a rich tooling environment for our users on ARCHER”.
CRESTA’s remit involves preparing applications and software for future exascale platforms, utilising a co-design approach. Beyond this, there is an extensive programme of research activities looking to develop new techniques and algorithms to exploit future exascale systems.
Our White Paper series is designed to showcase the social and economic importance of these activities and to highlight the benefits of the co-design approach, disseminating our work to the wider HPC community.
CRESTA is therefore pleased to announce the release of three new case studies and one new White Paper. The case studies showcase the work of the HemeLB, NEK5000 and Ludwig codes while the new White Paper looks at the latest developments in exascale architectures and discusses the implications for the CRESTA co-design applications.
The new case studies and White Paper are available from the Publications section of our website.
Copies of the slides and Presentations from the Joint Exascale Projects Workshop in Edinburgh
Professor Mark Parsons - Welcome
Panagiotis Tsarchopoulos - HPC in H2020 - Slides
Overview of the projects:
Wilfried Verachtert - EXA2CT - Slides
Stefano Markidis - EPIGRAM
Riccardo Rossi - NUMEXAS- Slides
Alex Ramirez - Mont-Blanc and Mont-Blanc 2- Slides
Estela Suarez - DEEP and DEEPER - Slides
Lorna Smith - CRESTA- Slides
Bernd Mohr - EESI-2 - Slides
Tools for Exascale - including new user requirements:
Bernd Mohr - DEEP and Mont-Blanc - Scalasca - Slides
Judith Giménez - DEEP and Mont-Blanc - Extrae/Paraver- Slides
Tobias Hilbrich - CRESTA - Vampir and CRESTA Tools - Slides
Stefano Markidis - EPiGRAM - Use Case of Tools in EPiGRAM - Slides
Riccardo Rossi - NUMEXAS - Slides
Eric Petit - EXA2CT
Programming models for Exascale:
Daniel Holmes - EPiGRAM - Message passing and PGAS - Slides
Jesus Labarta - Mont-Blanc (and DEEP) - OmpSs
Alistair Hart - CRESTA - OpenACC - Slides
Norbert Eicker - DEEP - ParaStation Global MPI
Eric Petit - EXA2CT
Innovative Algorithms for Exascale:
Wim Vanroose - EXA2CT - Slides
Riccardo Rossi - NUMEXAS
Dmitry Khabi - CRESTA - Slides
Pictures from the Collaboration Workshop that was held in Edinburgh on the 18 & 19th March 2014, thank you to everyone that attended.
Please email Andy (a.mcdonald at epic.ed.ac.uk) if you would like a hi-res version of any of the photos. Website file size limitations don't allow us to upload them here.
The CRESTA project has just published a new White Paper- The Exascale Development Environm…
UCL breaks through scaling roadblock with development tools at scale and achieves first si…